TWI757887B - Method, network controller, and computer program product for facilitating multipath transmission of a data stream from a sender to a receiver - Google Patents

Abstract

A method for facilitating multipath transmission of a data stream from a sender to a receiver, a network controller for a software-defined network comprising a plurality of switches for executing Packet forwarding from the sender to the receiver, the method comprising: performing operations at the network controller to cause the data stream from the sender to the receiver through multipath transmission using multiple network paths end, so that the multi-path transmission is transparent to the sender and the receiver. The network controller may be configured to facilitate the implementation of a multi-path random switching transport mechanism or a random switching network coding transport mechanism based on this method, thereby helping to improve transmission of a single data stream through multi-path transmission efficacy.

Description

Method, network controller, and computer program product for facilitating multipath transmission of a data stream from a sender to a receiver

The present invention relates to a transmission technology for multi-path transmission of data streams, and more particularly, to a method, network controller and computer program product for facilitating multi-path transmission of a data stream from a sender to a receiver.

In addition to text data, the data transmitted on the Internet now has a large amount of audio and video data. The transmission of audio and video data usually requires high bandwidth and immediacy. Therefore, improving the transmission efficiency of the network is a very important issue. However, when the packets are transmitted in the network, there may be packet loss (Packet Loss) due to various factors such as unstable line, load exceeding the available bandwidth or congestion, which will cause network transmission performance. drop; especially in the wireless network environment, due to changes in node mobility and wireless propagation (Propagation) conditions, the transmission path may be unstable due to signal source interference, obstructions, and other factors, resulting in unstable transmission quality, or even too long delay time This leads to packet loss, which seriously reduces transmission performance. Some studies have pointed out that in wireless networks, the packet loss probability of most connections is higher than 30%. Generally, a network in which packet loss occurs during packet transmission is called a lossy network. How to improve the data transmission performance in a lossy network is a very important issue.

As far as the existing network technology is concerned, the single data stream transmitted between the sender and the receiver is point-to-point transmission to the switch, and the routing path of the data stream is fixed, and the switch cannot perform multi-path transmission. Therefore, in order to develop higher network transmission performance, in recent years, whether in academia, industry or standard organizations (such as IETF, Internet Engineering Task Force) have studied how to use multi-path to transmit packets. Many transport mechanisms for multipath have been proposed today, such as using Equal Cost Multipath (ECMP) routing technology or IEEE 802.1aq link aggregation technology, which uses Mac address, IP address and TCP/UDP port to calculate data The hash value of the flow is used as the basis for adjusting the flow, so as to achieve the purpose of transmitting several data flows through different paths. In this way, it is necessary to find several transmission paths with equal cost (usually the same available bandwidth), and in fact, the packets in a data stream have the same source and sender, because the source of the packet When the IP and the destination IP are the same, according to the routing rules, they will be transmitted on the same path, so this type of technology will actually be difficult to achieve the effect of splitting and multiplexing the same data flow.

In addition, the IETF revised the TCP protocol and proposed the MPTCP (Multi-Path TCP) protocol, which improves the efficiency and reliability of network transmission through multiple paths. MPTCP is to open an additional subflow (Subflow) when establishing a TCP connection, disperse the data to multiple paths for transmission (still share the same transmit or receive buffer), by giving each connection a different connection Identifier, which reassembles the sub-data stream in the same connection into a single network stream, MPTCP protocol and sets a sequence number for each sub-data to detect loss and retransmit. Such a design actually combines multiple sub-data streams into one data stream, and the combined data stream can be said to be transmitted by multiple paths.

Ideally, MPTCP throughput (Throughput) should be as high as multiple disjoint single-path connections. However, in fact, when MPTCP transmits data, the total transmission volume is much lower than expected due to different packet loss rates and delay times of different transmission paths. Therefore, in the implementation of MPTCP, the performance of each transmission path must be considered first, so that the performance of each transmission path should not be too different, in order to achieve the expected benefits of multi-path transmission. In addition, the MPTCP mechanism operates on the third and fourth layer protocols, which are point-to-point (End to End) protocols, so the packets in one data stream (or sub-data stream) will still follow the same routing path (Routing). path), so in fact, MPTCP is a multi-data stream (or sub-data stream) for different fixed path transmission respectively.

In addition, in practice, the premise that the sender and receiver can use MPTCP is that each must have at least two network cards (such as Wi-Fi, 4G network cards) to provide at least two network addresses, And when the sender and the receiver decide to use MPTCP, the sender needs to implement the process of dispersing the data into multiple sub-data streams, and the receiver needs to realize the process of merging the multiple sub-data streams into original data. In this way, each of the sender and the receiver needs to install software that supports MPTCP or modify the original application to support MPTCP before use, and additionally increase the computing resources and hardware resource consumption of the sender and receiver.

Therefore, in order to improve the data transmission performance in the lossy network, the current transmission technology still needs to be improved.

One object of the present invention is to propose a technique for facilitating multipath transmission of a data stream from a sender to a receiver, suitable for a software-defined network (SDN), and the multipath transmission to the sender It is transparent to the receiving end. By implementing this technology in SDN, the sender and the receiver can benefit from the multi-path transmission of a single data stream according to the technology to improve transmission performance.

In order to achieve at least the above objects, the present invention provides a method for facilitating multipath transmission of a data stream from a sender to a receiver, for a network controller of a software-defined network (SDN) , the software-defined network includes a plurality of switches for performing packet forwarding from the sending end to the receiving end, the method comprising: performing a plurality of operations on the network controller to enable the transmission from the sending end The data stream is transmitted to the receiver through multi-path transmission using multiple network paths, so that the multi-path transmission is transparent to the sender and the receiver. The operations include: sending a first control message to a first switch of the switches, wherein the first switch receives packets of the data flow, the first control message including first data flow control information, The first data flow control information is used to notify the first switch to perform packet forwarding according to the individual path weights associated with the network paths; and to send a second switch to a second switch of the switches a control message, wherein the second switch receives packets corresponding to the data flow from the network paths, the second control message includes second data flow control information, and the second data flow control information is used to notify the first Two switches perform packet reception from the network paths and perform packet forwarding to the receiving end.

In one embodiment of the present invention, the operations further include obtaining network link statuses of the network paths from the SDN to determine at least the paths in the first data flow control information based on the network link statuses Weights.

In one embodiment of the present invention, the network link status includes at least one parameter group, the parameter group representing a corresponding network link attribute associated with the network paths individually, wherein the network link The attribute is one of multiple attributes including bandwidth, packet loss rate, delay, and jitter.

In an embodiment of the present invention, the at least one parameter group of the network link state includes a first parameter group and a second parameter group, the first parameter group representing the frequency respectively associated with the network paths. wide, the second parameter set represents the packet loss rate individually associated with the network paths.

In an embodiment of the present invention, the network controller is further configured to make the first control message further include data flow control information of a first network function, and the data flow control information of the first network function is used for informing the first switch to forward the data flow to at least one network node outside the network paths to perform the first network function.

In an embodiment of the present invention, the data flow control information of the first network function is further used to notify the first switch after receiving the adjusted packet corresponding to the first network function from the at least one network node. , so that the first switch performs packet forwarding according to the individual path weights associated with the network paths based on the first data flow control information.

In an embodiment of the present invention, the first network function includes network coding.

In an embodiment of the present invention, the network controller is further configured to make the second control message further include data flow control information of a second network function, and the data flow control information of the second network function is used for to notify the second switch to forward the received packet to at least another network node outside the network paths to perform the second network function after receiving the packet based on the second data flow control information.

In an embodiment of the present invention, the data flow control information of the second network function is further used to notify the second switch to receive an adjusted packet corresponding to the second network function from the at least another network node , so that the second switch performs packet forwarding to the receiving end based on the second data flow control information.

In one embodiment of the present invention, the second network function includes network decoding.

In one embodiment of the present invention, the operations further include: obtaining a representative network link state of the network paths from the SDN to determine a third network link state based on the representative network link state at least updated path weights in data flow control information; and sending the third control message to the first switch, wherein the third control message includes the third data flow control information, the first data flow control information for The first switch is instructed to perform packet forwarding according to the individually updated path weights associated with the network paths.

In one embodiment of the present invention, the representative network link state includes at least one parameter group, the parameter group representing a corresponding network link attribute individually associated with the network paths, wherein the network The link attribute is one of multiple attributes including bandwidth, packet loss rate, delay, and jitter.

In order to achieve at least the above object, the present invention further provides a computer program product, which includes a plurality of instructions, and the instructions are used to cause the computer to execute the process of facilitating a data flow from a sender to pass through as described in any one of the above-mentioned embodiments. A method for multipath transmission to a receiver.

In order to achieve at least the above objects, the present invention further provides a network controller for a software-defined network (SDN), the software-defined network includes a plurality of switches, and the switches are used for executing the slave For packet forwarding from a sender to a receiver, the network controller includes: a network interface unit and at least one processing unit. The network interface unit is configured to communicate with the SDN. the at least one processing unit, coupled to the network interface unit, is configured to perform a plurality of operations to cause a data stream from the sending end to be transmitted to the receiving end through multi-path transmission using a plurality of network paths, The multipath transmission is made transparent to the sender and the receiver. The operations include: sending a first control message to a first switch of the switches, wherein the first switch receives packets of the data flow, the first control message including first data flow control information, The first data flow control information is used to notify the first switch to perform packet forwarding according to the individual path weights associated with the network paths; and to send a second switch to a second switch of the switches a control message, wherein the second switch receives packets corresponding to the data flow from the network paths, the second control message includes second data flow control information, and the second data flow control information is used to notify the first Two switches perform packet reception from the network paths and perform packet forwarding to the receiving end.

In some embodiments of the network controller, the network controller may be used to implement at least one or any suitable combination of the various embodiments of the above methods.

The above-mentioned embodiments are based on SDN and apply the multi-path transmission of the second layer in the standard model of network communication protocol (for example, when transmitting packets of a single data stream in the network, the intermediate forwarding (Forward) node can pass through different transmission interfaces to forward different packets), so the multi-path transmission is transparent to the sender and the receiver. In this way, by implementing this technology in SDN, a single data stream can be transmitted through multiple paths to improve transmission performance, and the sender does not need to be additionally configured to support two network cards and the sender and receiver do not need to be installed. Specific software or modification of the original application program, so the technology has the advantage of convenience in practical application.

In order to fully understand the purpose, features and effects of the present invention, the present invention is described in detail by the following specific embodiments and in conjunction with the accompanying drawings, and the description is as follows:

Several embodiments are presented below to illustrate various implementations of the technique for facilitating multipath transmission of a data stream from a sender to a receiver according to the present invention. The technology is based on a software-defined network (SDN), and the multi-path transmission is transparent to the sender and the receiver. By implementing this technology in SDN, the sender and the receiver can benefit from the multi-path transmission of a single data stream according to the technology to improve transmission performance.

Please refer to FIG. 1 , which is a schematic diagram of an embodiment of a network architecture based on a software-defined network 20 to facilitate multipath transmission of a data stream from the sender 11 to the receiver 14 . As shown in FIG. 1 , the network architecture includes a software-defined network (SDN) 20 for facilitating multipath transmission of a data stream from the sender 11 to the receiver 14 , where, for example, the data stream It is an audio-visual (video), video or audio data stream, or a data stream of any data attribute, and the packets in the data stream have the same source and sender. The software-defined network 20 includes a plurality of switches supporting SDN, such as the switches 210 , 211_1 , 211_2 , and 220 shown in FIG. 1 , and the switches are used to perform packet forwarding from the sender 11 to the receiver 14 . These switches can be physical network switches or routers, and can also be implemented by computing devices with computing capabilities (such as computers or servers, etc.). The software-defined network 20 further includes one or more network controllers 200, and the network controllers 200 are used to perform a plurality of operations to control or configure the packet processing or forwarding behavior of the switches, so that the source The data stream from the sender 11 is multi-pathed to the receiver 14 using multiple network paths (eg, network paths P1, P2 as shown in FIG. 1 ), so that the multi-path transfers to the sender 14 The terminal 11 and the receiving terminal 14 are transparent, wherein the network path P1 refers to the packet passing through the switch 210, the switch 211_1, and the switch 220 to the receiving terminal 14, and the network path P2 refers to the packet From the switch 210 , the switch 211_2 , and the switch 220 to the receiving end 14 , and so on for the network path below. To this end, the network controller 200 may be configured to facilitate a method of multipathing a data stream from the sender to the receiver to facilitate the multipathing of a single data stream, which will be implemented based on FIG. 3 . Examples are described later. For example, due to the network transparency of the above-mentioned multi-path transmission for the sender and the receiver, in the application scenarios of SDN-based communication networks, such as computer networks, Internet, 4G, 5G Or any mobile communication network, or a hybrid communication network further combined with a fixed network (such as optical fiber communication or telecommunication network), the sender 11 and the receiver 14 can use the original network communication software, and it is not necessary to Any special configuration based on the path transmission mechanism can benefit from the multi-path transmission of a single data stream to improve transmission performance. In addition, the multi-path transmission is the multi-path transmission using the second layer (Layer 2) of the standard model of network protocols. For example, when transmitting a single data stream packet in the network, the intermediate forwarding (Forward) node can pass through Different transmission interfaces are used to forward different packets. For example, the network controller 200 can perform a plurality of operations to control or configure a switch (such as the switch 210 ) of the switches to perform stochastic switching packet processing or forwarding, so as to Passing the data stream from the sender 11 through the use of multiple network paths. The above-mentioned random switching on the multi-path transmission can be called a transmission mechanism of Multi-Path Stochastic Switching (MPSS), and the implementation method will be described with examples later.

Please refer to FIG. 2 , which is a schematic diagram of another embodiment of a network architecture based on a software-defined network 20 to facilitate multipath transmission of a data stream from the sender 11 to the receiver 14 . The embodiment of FIG. 2 is based on the embodiment of FIG. 1 , and may be further combined with a network function virtualization technology according to transmission requirements or network conditions. In FIG. 2 , the network controller 200 is used to perform multiple operations to control or configure the packet processing or forwarding behavior of the switches 210 , 211_1 ˜ 211_N and 220 , so that the The data stream is delivered to the receiver 14 through multi-path transmission using a plurality of network paths (such as network paths P1 to PN as shown in FIG. 1, where N is a positive integer greater than 1), such that the multi-path transmission pair The transmitting end 11 and the receiving end 14 are transparent. The network controller 200 can define a virtual network function (Virtual Network Function), and the shilling switch 210 forwards the packets of the data flow to another network node (such as the network node 12) outside the network paths ) to perform a network function, such as network coding. Correspondingly, the network controller 200 can make the switch 220 forward the packets from the network paths to another network node (such as the network node 13 ) outside the network paths to execute another The network function, for example, performs decoding corresponding to the aforementioned network code, and then forwards the decoded packet to the receiving end 14 . The network node is, for example, a computing device with computing and networking capabilities, such as a computer, a server, or other computing devices. The above-mentioned stochastic switching and network coding in multi-path transmission can be referred to as a transmission mechanism of Stochastic Switching and Network Coding (SSNC).

1 and 2 are used as examples to illustrate the network architecture based on software-defined network to facilitate a data stream to be transmitted from the sender to the receiver through multiple paths. However, the implementation of the present invention is not limited to the above examples. For example, in the example based on the network architecture of FIG. 1 or FIG. 2, each network path between switches 210 and 220 may have one or more switches, or no other switches; There may also be a switch between the terminal 11 and the switch 210, or between the switch 220 and the receiving terminal 14; the above-mentioned network architectures based on FIG. 1 or FIG. 2 are also suitable for implementing a multi-path random switching mechanism (MPSS) or Random Switched Network Coding (SSNC) transport mechanism.

Please refer to FIG. 3 , which is a schematic flowchart of one embodiment of a method for facilitating multipath transmission of a data stream from a sender to a receiver. The method of FIG. 3 is used to facilitate the multi-path transmission of a data stream from a sender 11 to a receiver 14 , and can be applied to network control based on a software-defined network (SDN) 20 as shown in FIG. 1 or FIG. 2 device 200. The method includes performing operations on the network controller 200 to cause the data stream originating from the sender 11 to be transmitted to the receiver 14 through multi-path transmission using a plurality of network paths such that the multi-path transmission pairs The transmitting end 11 and the receiving end 14 are transparent. As shown in FIG. 3 , the operations include operations as shown in blocks S110 to S120 , and FIG. 1 or FIG. 2 are used as examples when describing the method to facilitate understanding of its implementation, but it should be noted that the present invention Implementations are not limited to examples. The following method is performed under the condition that multiple network paths have been found.

As shown in block S110, the network controller 200 sends a first control message to one of the switches, wherein the switch directly or indirectly receives the packet of the data stream from the sender 11, The first control message includes first data flow control information for instructing the switch to perform packet forwarding according to the individual path weights associated with the network paths. For example, the network controller 200 can be used to execute the operation of block S110 to control or configure one of the switches including at least two or more packet transmission interfaces and to transmit packets to different links (Link). A switch, such as switch 210.

As shown in block S120, the network controller 200 sends a second control message to a switch of the switches, wherein the switch receives packets (eg, destination) corresponding to the data flow from the network paths the corresponding packets of the receiving end 14), the second control message includes second data flow control information, and the second data flow control information is used to notify the switch to perform packet reception from the network paths and to perform packet forwarding to the receiving end 14 . For example, the network controller 200 can be used to execute the operation of block S120 to control or configure the switches to include at least two or more packet transmission interfaces and to transmit packets from different links to the receiver 140 A switch, such as switch 220.

By implementing the above method in SDN, the multipath transmission is transparent to the sender and the receiver, and the sender and receiver can benefit from a single data stream through the multipath according to the method transmission to improve transmission performance.

Furthermore, as previously mentioned, various implementations are possible based on the network architecture of FIG. 1 or FIG. 2 , in which other switches than switches 210 , 220 may or may not exist. Therefore, for the network architecture in which other switches exist besides the switches 210 and 220, those skilled in the art to which the present invention pertains can, based on the method shown in FIG. In the related art of the channel, the network controller 200 is used to control other switches (eg, 211_1 to 211_N), so as to achieve multi-path transmission, such as multi-path random switching mechanism (MPSS) or random switching network coding (SSNC) transmission mechanism .

In addition, based on the method shown in FIG. 3 , other implementations can be further extended, such as processing or adjustment in response to transmission or network conditions, which will be exemplified below.

In block S110, the network controller 200 is used to communicate with the switch 210, so that the switch 210 can perform packet forwarding according to the individual path weights associated with the network paths, thereby enabling the switch 210 to perform random switching (stochastic switching) packet processing or forwarding behavior. In one embodiment, the network controller 200 can use the OpenFlow protocol to set the Group Table in the on-path switch 210 to the Select mode, and use multiple paths for packet forwarding at the same time. At this time, the network controller 200 can determine the path weight of the transmission path by setting the weight of the "Buckets" (ie, the Output Port) in the Select mode. For example, the goal of weight setting and adjustment is to make full use of the available bandwidth of the path. When a path failure or congestion occurs, because of the simultaneous transmission of multiple paths, it will not seriously affect the user, and the forwarding probability of the network path is allocated to other paths to restore traffic.

For example, the stochastic routing mechanism can be implemented in the network controller 200 using the Select type of the Group Table in the OpenFlow architecture, and it is called Stochastic Switching. The method is divided into two steps: (1) Provide the probability value of packet allocation for each output port. (2) Use an exchange to achieve random exchange. Step (2) is: calculating the routing path or assuming that the routing path is known, implement as far as possible to aggregate addresses of the same class into the same flow table entry (Flow Entry) in the data flow table (Flow Table). Because these data flows will have the same output port and output probability value, the network controller 200 will group the same group of data flows into the same group. The OpenFlow specification introduces the concept of "weight" for each bucket in the action bucket in the Group Table. The network controller 200 provides a weight value of each bucket (Bucket Weight) and a probability of specifying an output port. Each bucket uses the out_to_port action to forward packets to the output port.

The Select method of the Group Table in the aforementioned OpenFlow selects one of the buckets according to the weight value of the buckets to perform actions related to these buckets. Therefore, the implementation of the Select method may cause confusion first. number, and then select a corresponding bucket (Bucket) according to the random value, the algorithm is shown in the virtual code of Table 1, for example. Table 1 Procedure SELECT (m, w[]) //m=the number of buckets //w[i]=the weight of bucket i, 0＜=i ＜=m-1 Begin r = rand(); //generate a random number w = w[0]; for (i = 0 ; i <m; i++) { if ( r <= m ) return(i); //bucket i is selected else w = w + w[i+1 ]; } End

In one embodiment, it may be implemented using an Open vSwitch (OVS) that supports the OpenFlow specification. Open vSwitch (OVS) is an open source virtual switch based on software implementation, which can be integrated with many virtualization platforms and provides the capability of virtual switch Internet. Each OpenFlow switch has its own one or more OpenFlow data flow tables (Flow Table), and each data flow table will have multiple OpenFlow routing rules for processing packets, that is, data flow table entries (Flow Entry), each An OpenFlow routing rule is composed of fields such as matching fields, execution actions, and priorities. The match field contains the source port of the packet and the header of the packet, such as source IP address, destination port number, source MAC address, and so on.

The data flow table contains multiple so-called data flow table items (Flow Entry) data, each data flow table item (Flow Entry) includes the content that meets the conditions, the counter and the instruction to be issued for the matching network packet (Instruction). ) and so on, each data flow table entry (Flow Entry) corresponds to a data flow (Flow) transmitted in the network. Through the OpenFlow protocol, an OpenFlow controller (Controller) (such as the network controller 200 in FIG. 1 or FIG. 2 ) can add, modify and delete the contents of these data flow entries (Flow Entry). The OpenFlow Controller (Controller) is the backbone of SDN. Network administrators or developers can communicate with the OVS switch hardware through the open source code provided by the OpenFlow protocol, set the data flow table and define the transmission path (or routing path) of the packet. .

In addition, the Group Table allows each data stream to be assigned to a corresponding group, and can perform the same action for all data packets belonging to the same group identifier. There are four types of Group Table: All, Select, Indirect and Failover. The Select function selects one of multiple buckets according to the weight assigned to the bucket, and performs operations related to the bucket. Therefore, when the packet is transmitted, the port from which the packet is forwarded can be selected, and at the same time, the weight of the bucket can be matched, so that a data stream can be transmitted in multiple paths at the same time in a random manner.

In one embodiment, referring to the network architecture of FIG. 1 , the multi-path transmission data stream is forwarded by the sender 11 and the receiver 14 through the switch 210 , the switch 211_1 , and the switch 220 of the network path P1. The receiving end 14, wherein the probability of the switch 210 being forwarded to the switch 211_1 is 60%, and its link bandwidth is 30 Mbps; At the receiving end 14, the probability that the switch 210 transfers to the switch 211_2 is 40%, and its link bandwidth is 30 Mbps. For example, in the script example shown in Table 2 below, the above-mentioned probability of multi-path transmission and individual routing can be set by using the programming interface ovs-ofctl of the virtual switch of OVS. In the example in Table 2, a switch (eg switch 210) is configured using the programming interface ovs-ofctl and using parameters such as add-group (adding a group), group_id (group identifier), etc. , these parameters are defined by the ovs-ofctl programming interface of OVS. For details, please refer to the usage documents provided by the official website of the ovs-ofctl programming interface of OVS (such as the website http://www.openvswitch.org/support/dist-docs /ovs-ofctl.8.txt), which will not be repeated here. In Table 2, the parameter settings of "bucket=weight:60, output=2" can be used to set the probability that the switch 210 will forward the packet from the switch 210 to the switch 211_1 to 60%, "bucket=weight:40, The parameter setting of output=3" can be used to set the probability of the switch 210 to forward the packet from the switch 210 to the switch 211_2 to 40%. Table 2 ovs-ofctl -O OpenFlow13 add-group tcp:127.0.0.1:6673 group_id=1, type=select, bucket=weight:60, output=2, bucket=weight:40, output=3 ovs-ofctl -O OpenFlow13 add -flow tcp:127.0.0.1:6673 in_port=1,ip,nw_src=10.0.0.1,nw_dst=10.0.0.2,actions=group=1 ovs-ofctl -O OpenFlow13 add-flow tcp:127.0.0.1:6674 in_port= 1,ip,nw_src=10.0.0.1,nw_dst=10.0.0.2,actions=output=2 ovs-ofctl -O OpenFlow13 add-flow tcp:127.0.0.1:6675 in_port=1,ip,nw_src=10.0.0.1,nw_dst =10.0.0.2,actions=output=2 ovs-ofctl -O OpenFlow13 add-flow tcp:127.0.0.1:6676 in_port=1,ip,nw_src=10.0.0.1,nw_dst=10.0.0.2,actions=output=3 ovs -ofctl -O OpenFlow13 add-flow tcp:127.0.0.1:6676 in_port=2,ip,nw_src=10.0.0.1,nw_dst=10.0.0.2,actions=output=3

FIG. 4 is a schematic flow chart of one embodiment of the method of FIG. 3 . Compared with FIG. 3, the embodiment shown in FIG. 4 further includes a block S100. As shown in FIG. 4, the block S100 represents obtaining the network link status of the network paths from the SDN to be based on the network link The state determines at least the path weight in the first data flow control information. The embodiment of FIG. 4 shows that before the blocks S110 and S120 are performed, the network controller 200 can be used to detect the network link status of the SDN, so as to determine the path weight or further make other settings for the switch.

In one embodiment, the network link status includes at least one parameter group representing a corresponding network link attribute associated with the network paths individually, wherein the network link attribute includes One of the attributes of bandwidth, packet loss rate, delay, and jitter.

In one embodiment, the at least one parameter set of the network link state includes a first parameter set and a second parameter set, the first parameter set represents the bandwidths individually associated with the network paths, the The second parameter set represents the packet loss rates individually associated with the network paths.

。（公式1） In one embodiment, it is assumed that there are three network paths respectively P1, P2, and P3, the packet loss rates are L ₁ , L ₂ , and L ₃ (such as a first parameter group), and the available link bandwidths are B ₁ and B ₂ , B ₃ (such as a second parameter group), the path weights of each path are W ₁ , W ₂ , W ₃ , for example, based on the network architecture of Figure 1 or Figure 2, a single data stream is transmitted through P1, P2, and P3 (or network-encoded data stream as exemplified below). For example, the effective bandwidths of these paths would be E ₁ =B ₁ (1-L ₁ ), E ₂ =B ₂ (1-L ₂ ), E ₃ =B ₃ (1-L , respectively ₃ ), to transmit this data stream (or the data stream encoded by the network), it can be considered according to the effective bandwidth, so the path weights W ₁ , W ₂ , and W ₃ of these three paths can be set as:

. (Formula 1)

（公式2） 其中q=1、2或3。在一些實施例中，對於任意Q個網路路徑(Q＞1)，亦可參照公式2如此類推地推廣，從而確定各個網路路徑對應的路徑權重。此外，在一些實施例中，關於路徑權重的同一假設多條路徑的頻寬相同或相似的，則可基於封包的延遲(或抖動)的資料，參照與上述公式2相似的方式來確定各個網路路徑對應的路徑權重。在一些實施例中，可參照上述公式1或2相似方式，基於該網路鏈路屬性中一種、兩種或更多的屬性的資料來確定各個網路路徑對應的路徑權重。 In some embodiments, for any Q network paths (Q>1), formula 1 can also be used to generalize and so on, so as to determine the path weight corresponding to each network path. In another example, the path weight for each network path may be determined based on the packet loss rate, assuming the multiple paths have the same or similar bandwidth. For the previous embodiment, the path weights W ₁ , W ₂ , and W ₃ of the three network paths can be set as:

(Equation 2) where q=1, 2, or 3. In some embodiments, for any Q network paths (Q>1), formula 2 can also be used to generalize and so on, so as to determine the path weight corresponding to each network path. In addition, in some embodiments, it is assumed that the bandwidths of multiple paths are the same or similar with respect to the path weight, and then each network can be determined based on the packet delay (or jitter) information in a manner similar to the above formula 2. The path weight corresponding to the road path. In some embodiments, the path weight corresponding to each network path can be determined based on the data of one, two or more attributes of the network link attributes in a similar manner with reference to the above formula 1 or 2.

FIG. 5 is a schematic flowchart of an embodiment of the block S110 in FIG. 3 . As shown in block S111, the network controller 200 is configured to cause the first control message to include data flow control information of a first network function. As shown in block S115, the network controller 200 is configured to make the first control message include first data flow control information.

In one embodiment, the network controller 200 is further configured to make the first control message further include data flow control information of a first network function, and the data flow control information of the first network function is used to notify The switch 210 forwards the data stream to at least one network node outside the network paths to perform the first network function. As shown in FIG. 2 , the first network function is performed using a network node 12 . For example, the first network function includes network coding.

In one embodiment, the data flow control information of the first network function is further used to notify the switch 210 to receive an adjusted packet corresponding to the first network function from the at least one network node, thereby enabling the The switch 210 performs packet forwarding according to the individual path weights associated with the network paths based on the first data flow control information.

FIG. 6 is a schematic flowchart of an embodiment of the block S120 in FIG. 3 . As shown in block S121, the network controller 200 is configured to cause the second control message to include data flow control information of a second network function. As shown in block S125, the network controller 200 is configured to make the second control message include second data flow control information.

In one embodiment of the present invention, the network controller 200 is further configured to make the second control message further include data flow control information of a second network function, the data flow control information of the second network function It is used to notify the switch 220 to forward the received packets to at least another network node outside the network paths to perform the second network function after receiving the packets based on the second data flow control information. For example, in FIG. 2 , the second network function is performed using the network node execution 13 . For example, the second network function includes network decoding.

In an embodiment of the present invention, the data flow control information of the second network function is further used to notify the switch 220 to receive an adjusted packet corresponding to the second network function from the at least another network node, Therefore, the switch 220 performs packet forwarding to the receiving end 14 based on the second data flow control information.

In some embodiments, for example, based on the network architecture of FIG. 2 , a single data stream is transmitted through multiple paths P1 , P2 ˜PN in the network, wherein a multi-path random switching method is used as in the foregoing example, and the network node 12 is used. , 13 respectively perform the functions of network encoding and network decoding, so as to achieve the goal of improving transmission bandwidth and transmission efficiency. For example, based on the network architecture in FIG. 2 , the functions of network coding and network decoding are realized by using random linear network coding (Random Linear Network Coding, RLNC).

Regarding random linear network coding, for example, the switch 210 forwards the packet from the sender 11 to the network node 12, and the network node 12 implements an encoder (Encoder) 120 to perform the network coding operation, such as Use software programs, or a combination of software and hardware to achieve. Please refer to FIG. 7 , which illustrates an encoder implemented by the network node 12 to implement a network encoding method, such as random linear network encoding. As far as random linear network coding is concerned, it randomly selects k coefficients from the Finite Field to form a coding vector. coding. When the network node 12 receives the packets forwarded through the switch 210 and performs random linear network coding processing, the network node 12 divides a plurality of received packets (which can be regarded as a set of original packets) into several equal-sized packets. Block (Block), each block is subdivided into several symbols (Symbol). The network node 12 divides multiple packets into several blocks, for example, can be performed when multiple packets are received at one time, or can be performed when the number of received packets reaches a threshold value of one packet (eg 50, 100, 500 or 1000, etc.). Next, n encoding vectors are randomly generated to encode the symbols (Symbols) in the current block in a linear combination (Linear Combination) manner to generate n encoded symbols (Encoded Symbols), one of which is encoded by The product of a code vector and k symbols is summed up via an XOR operation. Assuming that the code rate (Code rate) is r, that is, after k symbols are coded by the network, n encoded symbols (Encoded Symbols) can be generated, then n=k*r. When the n encoded symbols are transmitted to the switch 220 through multi-path, the switch 220 forwards the n encoded symbols to the network node 13 for network decoding. The network node 13 implements a decoder (Decoder) 130 to perform the network decoding operation, for example, using a software program, or a combination of software and hardware. When receiving the coded symbols, the network node 13 will perform an XOR operation for decoding (Decoder). Assuming that the total number of coded symbols received by the network node 13 s>k, the network node 13 can convert the received coded symbols It is restored to the original symbol (which can be regarded as the aforementioned original packet set), which is then sent back to the switch 220 , and then forwarded to the receiving end 14 through the switch 220 . Although the encoded symbols (Encoded Symbols) may be lost during the transmission of multiple paths between the switch 210 and the switch 220, as long as the total number of received encoded symbols s>k, the sender does not need to The lost coded symbols are retransmitted, and the receiving end can restore the original symbols. Compared with the conventional single-path transmission method, the embodiment of the method proposed in the present invention utilizes multi-path transmission so that the transmission bandwidth can be increased, and the lost packets need not be retransmitted, thus the transmission efficiency can be improved.

In addition, the conventional method of transmitting the packets to the receiving end through a plurality of different paths may cause the problem of out-of-order of the received packets, but the embodiment of the method proposed in the present invention can overcome this problem, as described above In the embodiment shown in FIG. 7 , during the transmission process of multiple paths between the switch 210 and the switch 220, what is transmitted is the coded symbol, and the coded symbol has no sequence number, and the receiving end only needs to receive any sufficient number of coded symbols. The original symbols can be restored.

In addition, in this embodiment, the encoder and decoder of network coding are implemented in the network nodes 12 and 13 respectively. Therefore, both the multi-path transmission and the network coding are transparent to the sender and the receiver, and the sender 11 and the receiver 14 do not need to consume additional computing resources and hardware resources. In addition, although the random linear network coding is used as an example for illustration, the implementation of the present invention is not limited to the above-mentioned example. For example, with reference to the above-mentioned embodiment, any other suitable network coding method can be adjusted to be suitable for implementation in the network architecture based on FIG. 2 to become an embodiment of the present invention.

。 （公式3） 依據公式1將路徑權重代入上述公式，可以得出：

。 （公式4） 因為編碼符號總數量s大於未編碼的符號的數量k(即s＞k)，將此條件代入公式4，得出：

。（公式5） 藉此，在此實施例中，網路控制器200可以藉由網路鏈路狀態(例如，頻寬與封包遺失率)，得出一網路編碼的編碼率的下限值，如可以參考公式5來確定這個下限值，來控制該VNF以適當的編碼率來進行網路編碼。 In some embodiments, the aforementioned network coding is implemented by a virtual network function (VNF); in addition, a lower limit value of the coding rate of the network coding can be obtained according to the network link status , the network controller 200 can use this lower limit value to control the VNF to perform network encoding at an appropriate encoding rate, as shown below. For example, based on the network architecture of FIG. 2, for a single data stream, such as a video file or a video stream, that the sender 11 wants to transmit to the receiver 14, random switching and network coding are used for transmission, wherein the network node 12 is used to perform network encoding, and the network node 13 is used to perform corresponding network decoding. For other information about the switch, please refer to the relevant embodiments, so it will not be repeated. Assuming that the code rate of network coding is r, that is, if k symbols are subjected to network coding through the network node 12 to generate n coded symbols, then n=k×r, and the n coded symbols are transmitted through the network node 12. Switch 210 utilizes random switching and transmits through multiple network paths (eg, 3 network paths P1, P2, P3). After that, it is forwarded to the network node 13 through the switch 220 for network decoding, wherein the total number s of coded symbols received by the network node 13 should be:

. (Formula 3) Substitute the path weights into the above formula according to formula 1, we can get:

. (Equation 4) Since the total number s of encoded symbols is greater than the number k of unencoded symbols (ie, s>k), substitute this condition into Equation 4 to obtain:

. (Formula 5) In this way, in this embodiment, the network controller 200 can obtain a lower limit value of the coding rate of network coding according to the network link status (eg, bandwidth and packet loss rate) , for example, the lower limit value can be determined with reference to formula 5, so as to control the VNF to perform network coding at an appropriate coding rate.

FIG. 8 is a schematic flowchart of another embodiment of the method of FIG. 3 . Compared with the embodiment of FIG. 3 , the embodiment of FIG. 8 further includes blocks S130 and S140 . As shown in block S130, the network controller 200 obtains a representative network link state of the network paths from the SDN to determine a third data flow control based on the representative network link state The least updated path weight in the information. As shown in block S140, the network controller 200 sends the third control message to the switch 210, wherein the third control message includes the third data flow control information, and the first data flow control information is used to notify the Switch 210 performs packet forwarding according to the individually updated path weights associated with the network paths.

In one embodiment, the representative network link state includes at least one parameter group representing a corresponding network link attribute individually associated with the network paths, wherein the network link The attribute is one of multiple attributes including bandwidth, packet loss rate, delay, and jitter. As for the determination of the path weight, reference may be made to the foregoing related embodiments, or the embodiments related to Formula 1 or 2, etc. to implement.

Based on the embodiment shown in FIG. 8, the network controller 200 may be configured to continuously or dynamically monitor link conditions on the network. When it is found that there is a problem with the link or the link performance changes, such as an increase in the packet loss rate of the path or a decrease in the available bandwidth of the link due to path congestion, for the network architecture based on Figure 1 or Figure 2, the network control The controller 200 may be configured to recalculate the path weights based on Equation 1 or 2 (or a generalized formula thereof); or, for the network architecture based on FIG. 2, the network controller 200 may be configured to further check the encoding rate, such as using the lower limit value of the encoding rate based on Equation 5, to determine whether to change the encoding rate of network coding. In this way, the transmission performance of the above-mentioned multi-path transmission or the transmission performance of the multi-path transmission using network coding can be more effectively improved.

In some embodiments, the path weights may be dynamically adjusted. For example, it is assumed that at a certain point in time, the network controller 200 is configured to calculate the path weight based on Equation 1 or 2 (or its generalized formula), which is called the nominal weight, and the network controller 200 is configured as the nominal weight. 200 sends an instruction to the switch (eg, switch 210 ) through the OpenFlow specification to set the weight of the bucket selection in the group table, which is called the applied weight. In some dynamic network environments (such as wireless networks), changes in network traffic may be relatively large, so changes in link parameters (such as available bandwidth and packet loss rate) may be temporary. In some practical applications, in response to the above situation, the network controller 200 may perform some smoothing (Smoothing) processing on the calculated nominal weights to obtain practical application weights. For example, moving average is used for smoothing, which is calculated based on past values in a certain period of time. The calculated value can filter the noise of all values in the past period, which is helpful for Technical behavior to achieve numerical smoothing. The value calculated by this technique is an indicator that represents a trend following or staying behind. For example, in an example, if a moving average is to be calculated at a certain time point, a range of values (window) must be set. Suppose that a range of p intervals (such as p=20) is to be taken, then the calculated value range is The current parameters are added to the past p-1 parameter records, and then the p parameter records are averaged. The actual application weight value calculated in this way can more truly represent the change in the transmission performance of the current network link, whether good or bad. In another example, by smoothing the path weight or related processing technology, the network link performance data is further used to predict the future link performance trend, and blocks S130 and S140 in FIG. 8 are used to advance Allocating the transmission traffic of the path can avoid the congestion of the network link as soon as possible. In some embodiments, the network controller 200 may be configured to calculate the nominal weights at multiple time points through a moving average to obtain the actual application weights, and then issue an instruction to assign the actual application weights to the switches (eg, based on FIG. 1 or The switch 210 in FIG. 2 ) allows the switch to set the transmission probability of different transmission paths according to the actual application weight, so that the path weight can be dynamically adjusted. In this way, the multi-path transmission of a single data stream, such as the transmission mechanism of Multi-Path Random Switching (MPSS) or Random Switching Network Coding (SSNC), can be dynamically and adaptively performed efficiently.

9 is a schematic diagram of one embodiment of a network controller that may be used with SDN. As shown in FIG. 9 , the computing device 300 includes: a network interface unit 310 and at least one processing unit 320 . The computing device 300 can be used to implement the network controller 200 . The network interface unit 310 is configured to communicate with the SDN. The at least one processing unit 320, coupled to the network interface unit 310, is configured to perform multiple operations to enable a data stream from the sender 11 to arrive through multi-path transmission using multiple network paths The receiving end 14 makes the multi-path transmission transparent to the transmitting end 11 and the receiving end 14 . The operations include: sending a first control message to a switch 210 of the switches, wherein the switch 210 (eg, directly or indirectly from the sender 11 ) receives packets of the data stream, the first control message The message includes first data flow control information for instructing the switch 210 to perform packet forwarding according to the individual path weights associated with the network paths; and to one of the switches The switch 220 sends a second control message, wherein the switch 220 receives packets corresponding to the data stream from the network paths (eg, the corresponding packets destined for the receiving end 14 ), and the second control message includes the first Two data flow control information, the second data flow control information is used to notify the switch 220 to perform packet reception from the network paths and to perform packet forwarding to the receiving end 14 .

The computing device 300 may further include a memory unit 330 for storing the control logic 331 and the data 335 . The control logic 331 is, for example, code or instructions for implementing at least one embodiment of the present disclosure or a combination thereof (eg, the method embodiment based on any one of FIGS. 3 to 8 or a combination thereof). The data 335 is, for example, data that needs to be recorded or temporarily stored when the control logic 331 is executed, such as data such as control messages or network link status.

In some embodiments of the network controller 200, the network controller 200 may be used to implement at least one of the various embodiments of the above methods or any combination as appropriate.

Figure 10 is a schematic diagram of one embodiment of a switch that can be used with SDN. As shown in FIG. 10 , the switch 400 includes a packet forwarding unit 410 and a processing unit 420 . The switch 400 may be suitable for implementing a switch based on the network architecture in FIG. 1 or FIG. 2 . The packet forwarding unit 410 is connected to a plurality of ports 411_1 to 411_M (where M>1 is a positive integer). The switch 400 further includes a memory unit 430 for storing the control logic 431 and the data 435 . The control logic 431 is, for example, required for the corresponding switch 400 (such as the switch in FIG. 1 or FIG. 2 ) when implementing at least one embodiment (as shown in FIGS. 3 to 8 ) or a combination thereof based on the present disclosure. code or instructions. The data 435 is, for example, data that needs to be recorded or temporarily stored when the control logic 431 is executed, such as data flow control information or data to be used about the OpenFlow specification, such as a data flow table.

FIG. 11 is a schematic comparison diagram of the performance of a stream image using a conventional single-path transmission and the performance of the same stream image using a multi-path transmission according to an embodiment of the present invention. Single-path transmission in a general network environment can be established by finding the shortest path (Constrained Shortest Path, CSP) algorithm that meets certain conditions, which can establish a transmission path that meets certain conditions (such as the lowest bandwidth or delay, etc.), However, the established path is a single-path route. Therefore, once the link fails, the available bandwidth is insufficient, or the link is congested, the packet loss rate through the link will be greatly increased, and even the transmission path will be interrupted. It can be found from the experiment that when the available bandwidth of the link is insufficient, because the CSP mechanism is to establish a route of a single path, the link cannot switch paths, so the quality of streaming video transmission in the experiment is very poor. In Figure 10, the peak signal-to-noise ratio (PSNR) value corresponding to the single path transmission is observed in the representative curve C1, and it can be found that the transmission is very low from the beginning to the end, because There is no adjustment mechanism so there is no upward trend in quality. In addition, the average value of PSNR corresponding to curve C1 is, for example, 12.54dB.

According to an embodiment of the random switching network code transmission mechanism based on FIG. 2 , in the software-defined network environment, the random switching mechanism is used for transmission to achieve the effect of multiple paths of a single data stream. In this embodiment, the network is The controller monitors and records the link status of the network environment at any time, calculates the link weight of the data transmission at the same time, and combines the concept of moving average to achieve smooth weight adjustment, even if the link transmission performance drops during multi-path transmission , it can also adjust the link weight in time to improve the transmission quality. As shown in FIG. 10 , a representative curve C2 of the peak signal-to-noise ratio (PSNR) value corresponding to the multi-path transmission of the single data stream, wherein the average value of the PSNR corresponding to the curve C1 is, for example, 31.6dB. As shown in Figure 10, when the bandwidth is insufficient and the packet loss rate is too high after the 271st frame, according to the comparison results of the streaming video transmission experiment, this embodiment (represented by the curve C2 ) can effectively face the situation of insufficient available bandwidth and high packet loss rate in the network environment, and can properly allocate network traffic. Therefore, when the link transmission performance is poor, the use of the network path for transmission can be avoided immediately, which can not only avoid the problem of data transmission interruption due to the failure of a single transmission path, but also improve the utilization rate of network bandwidth and the overall The throughput of the network, the above experiments prove that it can achieve better transmission performance and streaming image transmission quality.

In some embodiments, a computer program product is provided that includes a plurality of instructions, and the instructions are used to cause a computer to execute the above-described method for facilitating multipath transmission of a data stream from a sender 11 to a receiver 14 at least one of multiple embodiments of . In some embodiments, the computer program product may include a storage medium, such as a non-transitory storage medium, storing instructions (or code) readable by a computing device, wherein the instructions are stored by at least one computing device (eg, The aforementioned network controller 200), when executed, enables the at least one computing device to implement at least one of the multiple embodiments of the aforementioned method. This method may be one or any combination of all the above-described embodiments including the method based on FIG. 3 . For example, the program code is, for example, one or more programs or program modules, such as for implementing blocks S110 to S120 based on FIG. 3 (or a method based on at least one embodiment of FIG. 3 to FIG. 8 or a combination thereof) , and can be executed in any suitable order. When a computing device (such as the aforementioned network controller 200 ) executes the code, it can cause the computing device to execute an embodiment of the operating method based on FIG. 3 . Examples of these storage media include, but are not limited to, optical information storage media, magnetic information storage media or memories such as memory cards, firmware or ROM or RAM. For example, the computing device includes a communication unit, a processing unit and a storage medium, wherein the processing unit is electrically coupled to the communication unit and the storage medium. The processing unit is used to communicate with the communication network in a wireless or wired manner through the communication unit, so as to communicate with other computing devices such as terminal devices. The processing unit may include one or more processors, and the computing device may also include other devices such as a graphics processor for performing operations. In some embodiments, the computing device may execute an operating system, and may further utilize various suitable network and software technologies such as web services, scripting engines, web applications, or web application programming interface (API) servers. achieved in one or more ways.

The above-mentioned embodiments are based on SDN and apply the multi-path transmission of the second layer in the standard model of network communication protocol (for example, when transmitting packets of a single data stream in the network, the intermediate forwarding (Forward) node can pass through different transmission interfaces to forward different packets), so the multi-path transmission is transparent to the sender and the receiver. With this network transparency, when the sender 11 wants to send the packet to the receiver 14, the sender address and the sender address (such as IP number, MAC, port) in the packet sent by the sender 11 do not need to be changed. So as to realize the multi-path transmission of a single data stream. In contrast, in the conventional MPTCP, multiple data streams (or sub-data streams) are transmitted on different fixed paths, and the sender must be additionally configured to support at least two network cards so that the multiple sub-data streams have their own different paths. The sender address (such as IP number, MAC at least one is different) and the sender and receiver also need to install specific software or modify the original application to achieve. Therefore, the embodiments of the present invention can promote the realization of multi-path transmission of a single data stream, and have the benefit of convenience in practical application for the sender and the receiver. In this way, by implementing this technology in SDN, it promotes the realization of the transmission mechanism of multi-path random switching or the transmission mechanism of random switching network coding, thereby helping to improve transmission performance through multi-path transmission of a single data stream.

The present invention has been disclosed above with preferred embodiments, but those skilled in the art should understand that the embodiments are only used to describe the present invention, and should not be construed as limiting the scope of the present invention. It should be noted that all the equivalent changes and substitutions to these embodiments should be considered to be included within the scope of the present invention. Therefore, the protection scope of the present invention should be defined by the scope of the patent application.

11: sender 12, 13: Network Nodes 14: Receiver 20: Software Defined Networking 120: Encoder 130: Decoder 200: Network Controller 210, 211_1, 211_2~211_N, 220: Switch 300: Computing Device 310: Network Interface Unit 320: Processing Unit 330: Memory Unit 331: Control Logic 335: Information 400: Exchanger 410: Packet forwarding unit 411_1~411_M: port 420: Processing Unit 430: Memory Unit 431: Control Logic 435: Information P1, P2~PN: network path S100, S110, S120: Blocks S111, S115, S121, S125: Blocks S140: Square C1, C2: Curve

FIG. 1 is a schematic diagram of one embodiment of a network architecture based on software-defined networking to facilitate multi-path transmission of a data stream from a sender to a receiver. FIG. 2 is a schematic diagram of another embodiment of a network architecture based on software-defined networking to facilitate multi-path transmission of a data stream from a sender to a receiver. 3 is a schematic flow diagram of one embodiment of a method for facilitating multipath transmission of a data stream from a sender to a receiver. FIG. 4 is a schematic flow chart of one embodiment of the method of FIG. 3 . FIG. 5 is a schematic flowchart of an embodiment of the block S110 in FIG. 3 . FIG. 6 is a schematic flowchart of an embodiment of the block S120 in FIG. 3 . FIG. 7 is a schematic diagram of an embodiment of implementing network coding based on the method of FIG. 3 . FIG. 8 is a schematic flowchart of another embodiment of the method of FIG. 3 . 9 is a schematic diagram of one embodiment of a network controller that may be used with SDN. Figure 10 is a schematic diagram of one embodiment of a switch that can be used with SDN. FIG. 11 is a schematic comparison diagram of the performance of a stream image using a conventional single-path transmission and the performance of the same stream image using a multi-path transmission according to an embodiment of the present invention.

11: sender

14: Receiver

20: Software Defined Networking

200: Network Controller

210, 211_1, 211_2, 220: Switches

P1, P2: network path

Claims (20)

A method for facilitating multipath transmission of a data stream from a sender to a receiver, for a network controller of a software-defined network (SDN), the software-defined network comprising a plurality of switches, the switches are configured to perform packet forwarding from the sender to the receiver, the method comprising: Performing operations at the network controller to cause the data stream from the sender to the receiver through multi-path transmission using multiple network paths such that the multi-path transmission to the sender and the receiver It is transparent, such operations include: sending a first control message to a first switch of the switches, wherein the first switch receives packets of the data flow, the first control message includes first data flow control information, the first data flow control information for instructing the first switch to perform packet forwarding according to the individual path weights associated with the network paths; and sending a second control message to a second one of the switches, wherein the second switch receives packets corresponding to the data stream from the network paths, the second control message includes second data Flow control information, the second data flow control information is used to notify the second switch to perform packet reception from the network paths and to perform packet forwarding to the receiving end. The method of claim 1, wherein the operations further comprise obtaining network link states of the network paths from the SDN to determine at least one of the first data flow control information based on the network link states path weight. The method of claim 2, wherein the network link status includes at least one parameter set representing a corresponding network link attribute individually associated with the network paths, wherein the network link The path attribute is one of multiple attributes including bandwidth, packet loss rate, delay, and jitter. The method of claim 3, wherein the at least one parameter set of the network link status includes a first parameter set and a second parameter set, the first parameter set representing the parameters associated with the network paths individually Bandwidth, the second parameter set represents the packet loss rate individually associated with the network paths. The method of claim 1, wherein the network controller causes the first control message to further include data flow control information of a first network function, and the data flow control information of the first network function is used to notify the The first switch forwards the data flow to at least one network node outside the network paths to perform the first network function. The method of claim 5, wherein the data flow control information of the first network function is further used to notify the first switch to receive an adjusted packet corresponding to the first network function from the at least one network node Then, the first switch is made to perform packet forwarding according to the individual path weights associated with the network paths based on the first data flow control information. The method of claim 6, wherein the first network function includes network coding. The method of claim 1, wherein the network controller causes the second control message to further include data flow control information of a second network function, and the data flow control information of the second network function is used to notify the second network function After performing packet reception based on the second data flow control information, the second switch forwards the received packet to at least another network node outside the network paths to perform the second network function. The method of claim 8, wherein the data flow control information of the second network function is further used to notify the second switch to receive the adjusted data corresponding to the second network function from the at least another network node packet, so that the second switch performs packet forwarding to the receiver based on the second data flow control information. The method of claim 9, wherein the second network function includes network decoding. The method of claim 1, wherein the operations further comprise: obtain a representative network link state of the network paths from the SDN to determine at least updated path weights in a third data flow control information based on the representative network link state; and sending the third control message to the first switch, wherein the third control message includes the third data flow control information, the first data flow control information is used to notify the first switch according to the network paths The associated individual updated path weights perform packet forwarding. The method of claim 11, wherein the representative network link state includes at least one parameter set representing a corresponding network link attribute individually associated with the network paths, wherein the The network link attribute is one of multiple attributes including bandwidth, packet loss rate, delay, and jitter. A computer program product comprising a plurality of instructions for causing at least one computing device to perform as described in any one of claims 1 to 12 to facilitate multi-path transmission of a data stream from a sender to a method on the receiving end. A network controller is used for a software-defined network (SDN), the software-defined network includes a plurality of switches, and the switches are used to perform packet forwarding from a sender to a receiver, The network controller includes: a network interface unit configured to communicate with the SDN; and at least one processing unit, coupled to the network interface unit, and configured to perform a plurality of operations to cause a data stream from the sending end to be transmitted to the receiving end through multi-path transmission using a plurality of network paths, such that The multipath transmission is transparent to the sender and the receiver, and the operations include: sending a first control message to a first switch of the switches, wherein the first switch receives packets of the data flow, the first control message includes first data flow control information, the first data flow control information for instructing the first switch to perform packet forwarding according to the individual path weights associated with the network paths; and sending a second control message to a second one of the switches, wherein the second switch receives packets corresponding to the data stream from the network paths, the second control message includes second data Flow control information, the second data flow control information is used to notify the second switch to perform packet reception from the network paths and to perform packet forwarding to the receiving end. The network controller of claim 14, wherein the operations further include obtaining network link statuses of the network paths from the SDN to determine the first data flow control information based on the network link statuses of at least the path weight. The network controller of claim 15, wherein the network link state includes at least one parameter group representing a corresponding network link attribute individually associated with the network paths, wherein the network link state The network link attribute is one of multiple attributes including bandwidth, packet loss rate, delay, and jitter. The network controller of claim 14, wherein the network controller is further configured to cause the first control message to further include data flow control information of a first network function, the data of the first network function The flow control information is used to notify the first switch to forward the data flow to at least one network node outside the network paths to perform the first network function, wherein the first network function includes network coding. The network controller of claim 17, wherein the data flow control information of the first network function is further used to notify the first switch to receive data corresponding to the first network function from the at least one network node After adjusting the packets, the first switch performs packet forwarding according to the individual path weights associated with the network paths based on the first data flow control information. The network controller of claim 14, wherein the network controller is further configured to cause the second control message to further include data flow control information of a second network function, the data of the second network function The flow control information is used to notify the second switch to forward the received packets to at least another network node outside the network paths to execute the second data flow control information after performing packet reception based on the second data flow control information A network function, wherein the second network function includes network decoding. The network controller of claim 19, wherein the data flow control information of the second network function is further used to notify the second switch to receive from the at least another network node corresponding to the second network function the adjusted packet, so that the second switch performs packet forwarding to the receiving end based on the second data flow control information.

Images